Computational models of cognitive deficits in Parkinson's disease
|Sebastien Helie and Erick J Paul (2015), Scholarpedia, 10(2):32137.||doi:10.4249/scholarpedia.32137||revision #147148 [link to/cite this article]|
What is Parkinson’s disease?
Parkinson’s disease (PD) is a degenerative disorder caused by the accelerated death of dopamine (DA) producing neurons. Within the substantia nigra pars compacta (SNpc), cell loss is predominately found in the ventral tier with less (but still extensive) damage in the dorsal tier (Fearnley and Lees 1991, Gibb and Lees 1991). In contrast, normal aging yields substantially less cell loss and in a dorsal-to-ventral pattern. Parkinsonian motor symptoms appear after a loss of 60-70% of SNpc cells and 70-80% of DA levels in striatal nuclei where SNpc cells send projections (Bernheimer et al. 1973, Gibb and Lees 1991). Motor symptoms include resting tremor, rigidity, bradykinesia (slow movements), and akinesia (loss of voluntary movements).
In addition to motor deficits, PD patients without dementia present cognitive symptoms that resemble those observed in patients with brain damage in the frontal lobes. Numerous studies documenting cognitive deficits of PD patients have revealed impairment in a variety of tasks related to memory, learning, visuospatial skills, and attention (Gotham et al. 1988). Notably, these deficits are not just observable in challenging and esoteric laboratory experiments, but they are also present in patients’ everyday lives (Poliakoff and Smit-Spark 2008); thus, a thorough understanding of cognitive deficits in PD is essential for improving patients’ quality of life.
The treatment of PD commonly involves pharmacological therapy aimed at improving motor symptoms. Many classes of drugs exist, but the most widely used are anticholinergics and dopaminergics. Anticholinergics block the uptake of acetylcholine (ACh), a neurotransmitter related to inhibition of movement via striatal nuclei (Benarroch 2012). Use of anticholinergics pre-dates dopaminergic drugs; however, dopaminergics have become the standard treatment for PD patients (Schapira 2005). Levodopa is the most commonly prescribed drug. It is a DA precursor—DA itself cannot cross the blood-brain barrier—that can help restore normal levels of DA in striatal nuclei thereby allaying motor symptoms. Dopamine agonists provide similar relief by acting directly on DA receptors. Unfortunately, these drug treatments can cause negative cognitive and neuropsychological side effects (Schapira 2005, Quinn 1995) and may sometimes be a predominant cause of cognitive deficits observed experimentally in PD (Price et al. 2009).
While there are a plethora of studies documenting cognitive deficits of PD patients (for a review, see Price et al. 2009), very few computational models have been proposed to investigate the variegated landscape of deficits observed in those studies. In this entry, we provide an overview of cognitive deficits in PD and review some computational models that account for those deficits.
Cognitive deficits in Parkinson’s disease
Patients suffering from Parkinson’s disease display many of the same deficits in executive functions as patients with frontal lobe damage (Owen et al. 1993). Executive functions demand attention, working memory, and logical reasoning to maximize performance. Evidence reviewed by Cools (2006) suggests that PD impairments in executive functions are all DA related. PD deficits in rule-learning (mediated by executive functions) have been well characterized in empirical investigations and include impaired rule-based category learning and rule maintenance, as well as increased perseverative response tendencies.
Ashby and his colleagues (2003) tested PD patients, age-matched controls, and younger controls in a rule-based categorization task similar to the Wisconsin Card Sorting Task (WCST) (Heaton et al. 1993). Like in the WCST, a simple one-dimensional rule could be used to categorize the stimuli perfectly. Compared to controls, significantly more PD patients failed to learn in this task than both the young and age-matched controls. More nuanced deficits have also been identified by using different kinds of rule-based tasks and performance metrics. For example, PD patients tend to demonstrate a failure of rule maintenance. Rule maintenance requires sustained attention to the relevant stimulus dimension (as determined by the rule) while ignoring variations in the other dimensions. Typically, rule maintenance is measured by set loss errors, which are defined as errors following several consecutively correct responses. In the WCST, PD patients exhibit significantly more set loss errors than controls (Beatty et al. 1989). Similarly, Filoteo, Maddox, Ing, and Song (2007) observed more set loss errors in a rule-based categorization task when the irrelevant dimensions varied randomly than when there was no variability in the irrelevant dimensions.
PD patients also appear to exhibit a perseverative tendency—patients often persist with the previous response strategy despite feedback suggesting a change in the relevant rule. Using a simplified version of the WCST (Nelson 1976), Gotham, Brown, and Marsden (1988) found PD patients to make significantly more perseverative errors than control participants. In addition, Beatty et al. (1989) found greater mean perseverative errors and responses than controls in the standard WCST. Finally, a meta-analysis of PD patient performances in WCST experiments found moderate effect sizes for both perseverative errors and perseverative responses, further supporting the hypothesis that PD patients exhibit perseverative tendencies (Zakzanis and Freedman 1999).
Research has also identified PD deficits in procedural learning, a kind of non-declarative memory (Squire 2004). Procedural learning is important in categorization tasks in which optimal responding cannot be obtained via logical reasoning or by using rule-based strategies. Shohamy and her colleagues (2005) reviewed evidence and collected data suggesting that at least some forms of procedural learning are DA-related. In a now classic study, Knowlton, Mangels, and Squire (1996) tested several patient groups in the Weather Prediction Task (WPT), a probabilistic classification task that requires participants to learn gradually to associate particular stimuli with the correct outcome. Knowlton and her colleagues found that PD patients performed significantly worse than controls in this task, and PD patients with the most severe symptoms never performed above chance.
Ashby and his colleagues (2003) tested PD patients in a categorization experiment in which the stimuli were separated into two categories in such a way that no easily verbalized rule would yield optimal performance. Interestingly, PD patients were unimpaired in this task compared to age-matched controls (although both groups were massively impaired relative to young controls). Similarly, PD patients showed no deficits in two non-verbal category-learning tasks that used two-dimensional continuous-valued stimuli when the categories were linearly separable, although they were impaired relative to controls when the categories were nonlinearly separable (Filoteo et al. 2005, Maddox and Filoteo 2001). These results suggest that PD patients are impaired relative to age-matched controls in tasks that rely on procedural learning, but only when the task is sufficiently complex. PD impairments are also evident in a procedural motor learning task whereby participants learn to execute a sequence of button presses (Pascual-Leone et al. 1993). The most widely used paradigm is known as the serial reaction time (SRT) task (Nissen and Bullemer 1987), wherein participants must quickly respond to the spatial location of visual stimuli by pressing keys corresponding to the spatial positions. Importantly, the appearance of stimuli in locations follows a fixed sequence that is unknown to the participant, but learned gradually over many repetitions. Though the exact nature of the deficit is debated (e.g., see Seidler et al. 2007), much evidence suggests that PD patients’ impairment is in procedures related to sequence learning (Siegert et al. 2006).
How is Parkinson’s disease typically modeled?
Because PD is characterized by the accelerated death of DA-producing neurons in the SNpc, PD is typically modeled by reducing the amount of DA available in computational models (Helie et al. 2013). Hence, in theory, any computational model that includes an explicit role for DA could be tested against data from PD patients to assess the role of DA in the model. Dopamine projections are predominantly directed at the basal ganglia (BG) and frontal cortex (Glimcher 2011), so many computational models of these brain structures include a role for DA. In addition, DA is thought to play an important role in BG-frontal functional connectivity by modulating the connectivity efficiency (Wickens and Kotter 1995). Computational models of cognitive deficits in PD have used one or both of these approaches (i.e., reduced DA levels and/or reduced BG-frontal connectivity) to simulate Parkinsonian symptoms.
Computational models of cognitive deficits in Parkinson’s disease
In this section, we briefly describe five computational models that have been used to simulate PD in chronological order. The reader is referred to the original papers for implementation details.
COVIS (Ashby et al. 1998) is a multiple-system model that was originally developed to account for the many behavioral dissociations between verbal and non-verbal categorization (Maddox and Ashby 2004). COVIS includes a hypothesis-testing system and a procedural learning system. The hypothesis-testing system can quickly learn a small set of (e.g., verbal) categories (those that can be found by hypothesis-testing and often can be verbally described) while the procedural learning system can learn any type of arbitrary categories in a slow trial-and-error manner (e.g., non-verbal). Each categorization system relies on a separate brain circuit, but they both include the BG. In the hypothesis-testing system, the BG is used to support working memory maintenance and for rule switching. In the procedural system, the BG is used to learn stimulus—response associations. Reducing DA levels in COVIS can account for many cognitive symptoms in PD patients such as perseveration, reduced sensitivity to negative feedback, and others (see Helie et al. 2012a, 2012b). While most COVIS simulations have used a rate version of the model, a spiking version of the procedural-learning system has been used to account for some categorization results and extended to account for instrumental conditioning (Ashby and Crossley 2011) and automaticity (Ashby et al. 2007). The spiking versions of COVIS could in theory also be used to simulate Parkinsonian symptoms.
Berns & Sejnowski (1998)
Berns and Sejnowski (1998) developed a computational model of sequence learning mediated by the BG and constrained by neuroanatomy. The critical feature of this model is that the BG learn sequences based on cortical states. To implement this process, the model includes a mechanism to select actions, and a mechanism to encode sequences within the BG. For action selection, the model assumes that: 1. the output nucleus of the BG selects among a set of possible actions and 2. the striatum represents cortical states and maps those states onto possible actions to be selected by the output nucleus (executed via the direct pathway of the BG). For sequence encoding, the model assumes that local recurrent connections within the indirect pathway of the BG represent a form of local feedback/working memory. Thus these recurrent projections in the indirect pathway learn a sequence of states represented by the striatum. The role of dopamine is modeled by the learning rate parameter used to update the connection strengths of the recurrent projections. Over many repetitions of the sequence, the model learns the sequence of states (via the internal recurrent projections) and reproduces the sequence with a single cue from the striatum. The model was used to account for PD performance in the SRT task by reducing the DA-dependent learning rate parameter, which evinced impaired sequence learning similar to PD (Pascual-Leone et al. 1993). The simulated symptoms displayed by the PD model include slower learning, longer simulated reaction time, and increased noise.
Monchi, Taylor, & Dagher (2000)
The Monchi et al. (2000) model was proposed to account for working memory deficits in PD and schizophrenia. The model includes three BG-thalamocortical closed loops: two with the prefrontal cortex (one for spatial information and the other for object information), and one through the anterior cingulate cortex (ACC). The role of the two prefrontal-BG loops is to maintain working memory information about the stimuli, whereas the ACC maintains the adopted strategy by inhibiting all the prefrontal cortex loops except one (i.e., representing the selected strategy). In the model, the visual stimulus is input to the prefrontal cortex loops, and the stimulus activity is propagated through the direct pathway of the BG. As a result, the thalamus is released from inhibition of the internal segment of the globus pallidus (GPi), and activation produced by the stimulus in the prefrontal cortex reverberates through closed-loops with the thalamus. When a response is required, the prefrontal cortex transfers its activation to the premotor cortex. If the response is incorrect, the nucleus accumbens sends a feedback signal to the ACC loop, which selects a new strategy by switching its inhibition to different prefrontal cortex loops. The Monchi et al. model has been used to simulate a delayed response task and the WCST. Interestingly, reducing the connection strengths within the BG-thalamocortical loops produces Parkinsonian symptoms, whereas reducing nucleus accumbens activity produces deficits similar to those observed in schizophrenia (Monchi et al. 2000).
The Frank (2005) model includes both the direct and indirect pathways through the BG, the premotor cortex, and an unspecified input area. In the Frank model, the input activates both the premotor cortex and the striatum. However, cortical activation is insufficient to produce a response, so BG processing is required to gate the correct response. The focus of the model is on: 1. the role of the indirect pathway in probabilistic learning and 2. the role of DA in probabilistic learning. In Frank’s model, the direct pathway is in charge of selecting the appropriate action (Go) whereas the indirect pathway is in charge of inhibiting inappropriate actions (NoGo). The direct and indirect pathways converge in the GPi and compete to control GPi activation, and eventually the response.
The competition between the direct and indirect pathways is modulated by DA. Specifically, higher DA levels increase activation in the direct pathway (e.g., through D1 receptors) and reduces activation in the indirect pathway (e.g., through D2 receptors). Hence, DA release following unexpected rewards results in long-term potentiation (LTP) in the direct pathway and long-term depression (LTD) in the indirect pathway. In contrast, DA dips following the unexpected absence of a reward reduces activation and produces LTD in the direct pathway but increases activation and produces LTP in the indirect pathway. The simulation results suggest that the dynamic range of the DA signal is crucial in probabilistic learning and reversal learning (e.g., when the response—reward associations are changed during learning). Reducing (to simulate PD) or increasing (to simulate medication overdose) DA levels can result in simulated Parkinsonian symptoms (Frank 2005).
Moustafa & Gluck (2011a, 2011b)
Moustafa and Gluck (2011a, 2011b) proposed a computational model of the striatum and prefrontal cortex that focuses on the DA projections to these areas as well as their interactions during multi-cue category learning. In the model, the prefrontal cortex is essential for attentional selection while the striatum is used for motor response selection. The role of DA in the model is to increase the gain of signaling and the learning rate. Simulating PD is done by reducing DA in the simulated prefrontal cortex and BG, which reduces the gain and learning rate in both these brain areas. Using this manipulation, the Moustafa and Gluck model can account for categorization deficits in PD patients.
Models of dopaminergic drugs
Two of the computational models described in the section on Computational models of cognitive deficits in Parkinson’s disease have been used to account for dopaminergic medication. Below we provide a mechanistic explanation. The reader is referred to the original papers for exact parameter settings.
In Frank (2005), the effect of dopaminergic medication is simulated by increasing DA levels in the model. This process can be used not only to restore the model’s performance by treating the simulated PD, but also to simulate cognitive deficits that are caused by dopaminergic medication. This is accomplished by implementing ideas from the overdose hypothesis (Gotham et al. 1988). According to the overdose hypothesis, the amount of medication required to replace DA in the dorsal striatum (and reduce motor symptoms) might ‘overdose’ the ventral striatum (which is mostly spared in early PD). The ventral striatum plays an important role in feedback processing, so overdosing this structure may create new cognitive deficits. In the Frank (2005) model, the overdose hypothesis is simulated by increasing the DA levels above baseline (intact) levels. This manipulation produces deficits similar to administering dopaminergic drugs in early PD in the probabilistic reversal learning task.
Moustafa & Gluck (2011a, 2011b)
In the Moustafa and Gluck (2011a) model, the effect of dopaminergic drug therapy is to increase tonic DA levels in both the PFC and the BG, which in turn reduces the range of phasic DA. This allows for simulating many manipulations in which the same participants were tested both ON and OFF medication. As reviewed in Frank (2005), PD patients ON medication sometimes do worse than participants OFF medication. For example, dopaminergic medication impairs performance of simulated PD patients in multi-cue probabilistic category learning. Additionally, the Moustafa and Gluck (2011b) model can account for some effects of anticholinergic medication on PD. The Moustafa & Gluck model assumes that the administration of anticholinergic medications in PD affects hippocampal function (Ehrt et al. 2010), which interferes with transfer generalization (Herzallah et al. 2010). This is simulated by removing the hippocampus from the model, which reduces the model’s performance in transfer generalization (Moustafa and Gluck 2011b).
Conclusion and Future Directions
This entry reviewed five computational models that have been used to reproduce cognitive symptoms related to PD. As mentioned in the section How is Parkinson’s disease typically modeled?, most of these models focus on the BG, and their interaction with prefrontal cortex. These brain structures have received much attention from computational modellers. In particular, some computational models of the BG are very anatomically detailed (e.g., Ponzi and Wickens 2010), and in theory any computational model of the prefrontal cortex or the BG that includes a role for DA could be used to account for PD. Hence, computational research focused on the BG and the prefrontal cortex could be symbiotic to research focused on cognitive deficits of PD. Specifically, research focused on cognitive deficits in PD provide for additional data that should be used to test computational models of the BG and prefrontal cortex. Likewise, computational models of the BG and prefrontal cortex can be used to generate new predictions and possibly new treatment possibilities. One of the goals of this entry is to provide a bridge to stimulate this relationship.
Although DA is the primary target of both therapy and research in PD, myriad neurotransmitters and neurotransmitter systems are compromised to varying degrees. For example, DA, norepinephrine, and epinephrine are all catecholamines, with the latter two synthesized from DA (Joh and Hwang 1987). Thus, disruption of DA production may lead to reductions in norepinephrine and epinephrine (since DA is the precursor). Additionally, sites of neurotransmitter production other than just the SNpc may be compromised in PD, leading to reductions in serotonin, norepinephrine and epinephrine (Halliday et al. 1990). Finally, neurotransmitter systems must be considered together to understand the impact of depletion of even one neurotransmitter. In the striatum, there is a critical balance between ACh and DA, which is thrown out of equilibrium in PD and may account for much of the striatal dysfunction evident in the disease (Aosaki, Miura, Suzuki, Nishimura, and Masuda 2010). It is likely that DA depletion is just one piece of a constellation of deficiencies leading to cognitive impairment in the disease. More sophisticated, biologically constrained computational models incorporating additional neurotransmitter systems may help to further elaborate current understanding and treatment of PD.
- Aosaki, T.; Miura, M.; Suzuki, T.; Nishimura, K. and Masuda, M. (2010). Acetylcholine-dopamine balance hypothesis in the striatum: An update. Geriatrics & Gerontology International 10: S148-S157.
- Ashby, F. G.; Alfonso-Reese, L. A.; Turken, A. U. and Waldron, E. M. (1998). A neuropsychological theory of multiple systems in category learning. Psychological Review 105: 442-481.
- Ashby, F. G. and Crossley, M. J. (2011). A computational model of how cholinergic interneurons protect striatal-dependent learning. Journal of Cognitive Neuroscience 23: 1549-1566.
- Ashby, F. G.; Ennis, J. M. and Spiering, B. J. (2007). A neurobiological theory of automaticity in perceptual categorization. Psychological Review 114: 632-656.
- Ashby, F. G.; Noble, S.; Filoteo, J. V.; Waldron, E. M. and Ell, S. W. (2003). Category learning deficits in Parkinson's disease. Neuropsychology 17: 115-124.
- Beatty, W. W.; Staton, R. D.; Weir, W. S.; Monson, N. and Whitaker, H. A. (1989). Cognitive Disturbances in Parkinsonʼs Disease. Journal of Geriatric Psychiatry and Neurology 2: 22-33.
- Benarroch, E. E. (2012). Effects of acetylcholine in the striatum. Neurology 79: 274-281.
- Bernheimer, H.; Birkmayer, W.; Hornykiewicz, O.; Jellinger, K. and Seitelberger, F. (1973). Brain dopamine and the syndromes of Parkinson and Huntington Clinical, morphological and neurochemical correlations. Journal of the Neurological Sciences 20: 415-455.
- Berns, G. S. and Sejnowski, T. J. (1998). A computational model of how the basal ganglia produce sequences. Journal of Cognitive Neuroscience 10: 108-121.
- Cools, R. (2006). Dopaminergic modulation of cognitive function-implications for L-DOPA treatment in Parkinson’s disease. Neuroscience and Biobehavioral Reviews 30: 1-23.
- Ehrt, U.; Broich, K.; Larsen, J. P.; Ballard, C. and Aarsland, D. (2010). Use of drugs with anticholinergic effect and impact on cognition in Parkinson's disease: A cohort study. Journal of Neurology, Neurosurgery and Psychiatry 81: 160-165.
- Fearnley, J. M. and Lees, A. J. (1991). Ageing and Parkinsonʼs disease: substantia nigra regional selectivity. Brain 114: 2283-2301.
- Filoteo, J. V.; Maddox, W. T.; Ing, A. D. and Song, D. D. (2007). Characterizing rule-based category learning deficits in patients with Parkinsonʼs disease. Neuropsychologia 45: 305-320.
- Filoteo, J. V.; Maddox, W. T.; Salmon, D. P. and Song, D. D. (2005). Information-integration category learning in patients with striatal dysfunction. Neuropsychology 19: 212-222.
- Frank, J. V. (2005). Dynamic dopamine modulation in the basal ganglia: A neurocomputational account of cognitive deficits in medicated and nonmedicated Parkinsonism. Journal of Cognitive Neuroscience 17: 51-72.
- Gibb, W. R. and Lees, A. J. (1991). Anatomy, pigmentation, ventral and dorsal subpopulations of the substantia nigra, and differential cell death in Parkinson’s disease. Journal of Neurology, Neurosurgery, and Psychiatry 54: 388-396.
- Glimcher, P. W. (2011). Understanding dopamine and reinforcement learning: the dopamine reward prediction error hypothesis. Proceedings of the National Academy of Sciences 108: 15647–15654.
- Gotham, A. M.; Brown, R. G. and Marsden, C. D. (1988). "Frontal" cognitive function in patients with Parkinson’s disease "on" and "off" levodopa. Brain 111: 299-321.
- Halliday, G. M.; Li, Y. W.; Blumbergs, P. C.; Joh, T. H.; Cotton, R. G. H.; Howe, P. R. C.; Blessing, W. W. and Geffen, L. B. (1990). Neuropathology of immunohistochemically identified brainstem neurons in Parkinson's disease. Annals of Neurology 27: 373-385.
- Heaton, R. K.; Chelune, G. J.; Talley, J. L.; Kay, G. G. and Curtiss, G. (1993). Wisconsin Card Sorting Test manual. Odessa, FL: Psychological Assessment Resources, Inc..
- Helie, S.; Paul, E. J. and Ashby, F. G. (2012a). A neurocomputational account of cognitive deficits in Parkinson's disease. Neuropsychologia 50: 2290-2302.
- Helie, S.; Paul, E. J. and Ashby, F. G. (2012b). Simulating the effect of dopamine imbalance on cognition: From positive affect to Parkinson's disease. Neural Networks 32: 74-85.
- Helie, S.; Chakravarthy, S. and Moustafa, A. A. (2013). Exploring the cognitive and motor functions of the basal ganglia: An integrative review of computational cognitive neuroscience models. Frontiers in Computational Neuroscience 7: 174.
- Herzallah, M. M.; Moustafa, A. A.; Misk, A. J.; Al-Dweib, L. H.; Abdelrazeq, S. A.; Myers, C. E. and Gluck, M. A. (2010). Depression impairs learning whereas anticholinergics impair transfer generalization in Parkinson patients tested on dopaminergic medications. Cognitive & Behavioral Neurology 23: 98-105.
- Joh, T. H. and Hwang, O. (1987). Dopamine β-Hydroxylase: Biochemistry and Molecular Biology. Annals of the New York Academy of Sciences 493: 342-350.
- Knowlton, B. J.; Mangels, J. A. and Squire, L. R. (1996). A neostriatal habit learning system in humans. Science 273: 1399-1402.
- Maddox, W. T. and Ashby, F. G. (2004). Dissociating explicit and procedural-learning based systems of perceptual category learning. Behavioural Processes 66: 309-332.
- Maddox, W. T. and Filoteo, J. V. (2001). Striatal contributions to category learning: Quantitative modeling of simple linear and complex nonlinear rule learning in patients with Parkinsonʼs disease. Journal of the International Neuropsychological Society 7: 710–727.
- Meco, G.; Casacchia, M.; Lazzari, R.; Franzese, A.; Castellana, F.; Carta, A.; Iannuccelli, M. and Agnoli, A. (1984). Mental impairment in Parkinson's disease. The role of anticholinergic drugs. Acta Psychiatrica Belgica 84: 325-335.
- Monchi, O.; Taylor, J. G. and Dagher, A. (2000). A neural model of working memory processes in normal subjects, Parkinson’s disease and schizophrenia for fMRI design and predictions. Neural Networks 13: 953-973.
- Moustafa, A. A. and Gluck, M. A. (2011a). A neurocomputational model of dopamine and prefrontal-striatal interactions during multi-cue category learning by Parkinson's patients. Journal of Cognitive Neuroscience 23: 151-167.
- Moustafa, A. A. and Gluck, M. A. (2011b). Computational cognitive models of prefrontal-striatal-hippocampal interactions in Parkinson's disease and schizophrenia. Neural Networks 24: 575-591.
- Nissen, M. J. and Bullemer, P. (1987). Attentional requirements of learning: evidence from performance measures. Cognitive Psychology 19: 1-32.
- Nelson, H. E. (1976). A modiﬁed card sorting test sensitive to frontal lobe defects. Cortex 12: 313-324.
- Owen, A. M.; Roberts, A. C.; Hodges, J. R. and Robbins, T. W. (1993). Contrasting mechanisms of impaired attentional set-shifting in patients with frontal lobe damage or Parkinson's disease. Brain 116: 1159-1175.
- Pascual-Leone, A.; Grafman, J.; Clark, K.; Stewart, M.; Massaquoi, S.; Lou, J.-S. and Hallett, M. (1993). Procedural learning in Parkinson's disease and cerebellar degeneration. Annals of Neurology 34: 594-602.
- Poliakoff, E. and Smith-Spark, J. H. (2008). Everyday cognitive failures and memory problems in Parkinson’s patients without dementia. Brain and Cognition 67: 340-350.
- Pondal, M.; Del Ser, T. and Bermejo, F. (1996). Anticholinergic therapy and dementia in patients with Parkinson's disease. Journal of Neurology 243: 543-546.
- Ponzi, A. and Wickens, J. (2010). Sequentially switching cell assemblies in random inhibitory networks of spiking neurons in the striatum. Journal of Neuroscience 30: 5894-5911.
- Price, A.; Filoteo, J. V. and Maddox, W. T. (2009). Rule-based category learning in patients with Parkinson’s disease. Neuropsychologia 47: 1213-1226.
- Quinn, N. (1995). Drug treatments of Parkinson’s disease. British Medical Journal 310: 575-579.
- Schapira, A. H. V. (2005). Present and future drug treatment for Parkinson’s disease. Journal of Neurology, Neurosurgery, and Psychiatry 76: 1472-1478.
- Seidler, R. D.; Tuite, P. and Ashe, J. (2007). Selective impairments in implicit learning in Parkinson’s disease. Brain Research 1137: 104-110.
- Shohamy, D.; Myers, C. E.; Grossman, S.; Sage, J. and Gluck, M. A. (2005). The role of dopamine in cognitive sequence learning: evidence from Parkinson’s disease. Behavioural Brain Research 156: 191-199.
- Siegert, R. J.; Taylor, K. D.; Weatherall, M.; Abernethy, D. A. and Gluck, M. A. (2006). Is implicit sequence learning impaired in Parkinson's disease? A meta-analysis. Neuropsychology 20: 490-495.
- Squire, L. R. (2004). Memory systems of the brain: A brief history and current perspective. Neurobiology of Learning and Memory 82: 171-177.
- Wickens, J. and Kotter, R. (1995). Cellular models of reinforcement. In J. C. Houks, J. L. Davis, & D. G. Beiser (eds.). Models of Information Processing in the Basal Ganglia (pp. 187-214) (Cambridge, MA: MIT Press).
- Zakzanis, K. and Freedman, M. (2006). A neuropsychological comparison of demented and nondemented patients with Parkinsonʼs disease. Applied Neuropsychology 6: 129-146.